Skip to main content

MIT Quest AI Roundtable

Extending Deep Nets to New,
Unexpected Situations

February 11, 2021 | 7pm - 8pm EST
Webinar

Extending Deep Nets to New, Unexpected Situations

Deep neural networks could very well memorize their training data, but instead they find generalizable rules.We will discuss various ideas for why this happens, and how we can build deep learning systems that generalize even better to new and unexpected scenarios

 

 

Speakers

Photo of Phillip Isola

Phillip Isola

Speaker
Bonnie & Marty (1964) Tenenbaum Career Development Assistant Professor, Department of Electrical Engineering and Computer Science
Computer Science and Artificial Intelligence Laboratory
photo of Pulkit Agrawal

Pulkit Agrawal

Speaker
Steven G (1968) and Renee Finn CD Assistant Professor, Department of Electrical Engineering and Computer Science
Computer Science and Artificial Intelligence Laboratory
Laboratory for Information and Decision Systems
photo of Alyosha Efros

Alyosha Efros

Professor
Department of Electrical Engineering and Computer Sciences
UC Berkeley
Reasons to love GANs
Photo of Aude Oliva

Aude Oliva

Moderator
Director, MIT Quest Corporate
MIT Director, MIT-IBM Watson AI Lab
Senior Research Scientist, MIT
Schwarzman College of Computing

Schedule

Schedule

Date: Thursday, February 11, 2021
Time: 7pm - 8pm EST
Where: Zoom Webinar

7:00 PM - 7:05 PM

Introduction
Aude Oliva

7:05 PM - 7:25 PM

Why do Deep Nets Generalize?
Phillip Isola

7:25 PM - 7:40 PM

"Unwanted" Generalization
Pulkit Agrawal

7:40 PM - 8:00 PM

Roundtable Discussion and Q&A
Phillip Isola, Pulkit Agrawal, Alyosha Efros and Aude Oliva